TILLING Project Planning in Soybean and Chickpea: Target Genes, Population Screening, and Sequence Validation

A TILLING project succeeds when the biological question, mutant resource, target design, pooling scheme, sequence screen, and seed inventory form one recoverable chain. Soybean projects must distinguish the intended locus from duplicated genes and close family members. Chickpea projects may first need to establish whether the available generation, DNA collection, and seed stocks constitute a screenable population. The plan should therefore begin with the evidence required from a confirmed line, work backward to the individual plant, and only then select pool geometry and sequencing scale.

Key takeaways

• Define the desired allele class and downstream experiment before choosing targets or sequencing depth.
• Treat DNA, line identity, plant generation, and retrievable seed as one linked resource.
• Design soybean assays against the project genotype and test paralog specificity before full screening.
• Normalize DNA and choose a pool design that preserves unambiguous candidate retrieval.
• Confirm candidates in individual lines with an independent aliquot or extraction before biological interpretation.
• Report what the screen could not test, including uncallable bases, missing lines, and unresolved loci.

Start With the Biological Question

The first planning question is not "Which platform should we use?" It is "What result would justify recovering a line?" A project seeking any protein-truncating allele in one gene has a different target space from a project seeking a graded missense series across a catalytic domain, promoter variants, or independent alleles in duplicated soybean genes. Write the desired evidence state before selecting an amplicon: candidate in a pool, candidate assigned to an individual, sequence-confirmed allele, viable seed stock, or segregating material ready for phenotyping.

TILLING searches a mutagenized population for induced variants; it does not place a predetermined nucleotide change at a chosen position. It can identify alleles within callable target regions, but it cannot promise that the population contains a specific substitution or that a detected variant changes a trait. This distinction prevents a technically successful screen from being judged against an unanswerable question.

Estimate discovery opportunity from the population’s measured mutation density, the number of screened individuals, and the number of callable target bases. These components scale the expected number of alleles; none should be replaced by a generic population-size rule. If mutation density is unknown, a pilot across representative loci can provide an empirical basis for expanding the screen. A project with several genes should rank them by biological value, targetability, and the consequence classes that would be useful rather than treating every base as equally informative.

Audit the Mutant Population

A collection becomes a screening population only when its records allow a DNA result to lead back to biological material. Record the founder genotype, mutagen and treatment, M1 family structure, sampled generation, selfing or advancement history, phenotype notes, DNA extraction batch, and seed inventory. M2 plants are commonly useful for detecting segregating induced variants, while later generations may offer more stable genotypes; however, generation labels are meaningful only if single-plant descent and seed handling are documented.

For a proposed chickpea population, decide whether the current material is ready for target screening or still in population development. The 2022 chickpea protocol by Amri-Tiliouine and colleagues links mutation induction, M2 assessment, DNA extraction, and mutation discovery as separate but connected stages. A shipment of M1 seed, an unindexed bulk, or M2 plants without paired seed cannot be treated like a mature DNA-and-seed library. Population advancement, viability testing, and inventory repair may be the true critical path.

Audit the following before target design:

• one stable line identifier shared by field book, DNA plate, sequence file, and seed packet;
• founder genotype and a reference DNA sample from the same genetic background;
• DNA concentration method, integrity observations, storage history, and remaining volume;
• plate maps, blanks, duplicates, missing wells, swaps, and any re-arraying events;
• generation and family relationships for every sampled plant;
• seed quantity, storage condition, viability information, and rules for distribution;
• measured mutation spectrum or density, if available, with the method used to estimate it;
• permission to consume DNA, re-extract tissue, advance seed, and share resulting data.

Quarantine identity conflicts rather than forcing them into a pool. Missing DNA can reduce effective population size, but an incorrect DNA-to-seed link creates a more serious failure: the project may confirm a real mutation in a line that cannot be recovered. The related T-DNA insertion analysis resource provides a useful contrast between insertion-defined material and chemically mutagenized populations, where target mutations are distributed across the genome and line-level traceability must carry the discovery process.

Choose Targets Before Pooling

Target design converts a gene name into testable sequence. Freeze the reference assembly version, gene model, transcript, strand, exon coordinates, and project-genotype sequence. If the founder differs from the public reference, those differences can disrupt primer sites or appear as recurrent non-induced variants. Reviewing existing soybean variation and the soybean SNP detection resource helps separate standing polymorphism from the induced-allele question, but the founder control remains the most relevant comparator.

Soybean’s duplicated genome makes specificity a primary design constraint. Cooper and colleagues found that many early primer sets amplified multiple products even when gel size appeared acceptable. A single visible band is therefore insufficient evidence of locus specificity. Align the target against paralogs and gene-family members, place discriminatory bases near primer ends when possible, and confirm the pilot product by sequence. If a locus-specific amplicon cannot be built, redesign the region or explicitly adopt an analysis capable of separating copies; do not silently merge signals.

For each amplicon, retain the primer sequences, expected product, covered genomic coordinates, transcript coordinates, intended consequence classes, paralog checks, pilot result, and excluded bases. The soybean genome sequencing resource can help teams frame reference and resequencing inputs when the starting genotype is poorly represented by a public assembly.

Design Pools for Retrieval

Pooling saves library and sequencing capacity only when each candidate can be deconvoluted. In a one-dimensional design, a positive pool must usually be reopened and its members retested. Two-dimensional designs place each sample into intersecting pools so that matched signals identify a candidate coordinate. Three-dimensional or indexed schemes can increase efficiency, but they also increase dependence on precise liquid handling, balanced DNA input, unique pool membership, and software that understands the design.

Choose geometry from the number of samples, expected mutation density, assay noise, sequencing error, DNA volume, target count, and acceptable confirmation burden. Deeper pools reduce per-sample cost but dilute an induced allele and can make low-quality members invisible. Shallower pools increase library count but simplify candidate assignment. Model expected allele fraction for heterozygous and homozygous states, then test the poorest acceptable DNA class rather than only high-quality controls.

DNA normalization is part of variant sensitivity, not a clerical step. Quantify samples with a method suited to amplifiable DNA, define the allowed concentration range, document dilutions, and flag low-volume or inhibited wells. Include a founder control, no-template controls, known variants where available, and technical duplicates that pass through pooling. Plate and pool IDs should be immutable. A pool manifest must state every member, contribution, source plate position, transfer date, operator or automation run, and any exception.

Run a small end-to-end pilot before constructing all pools. It should exercise target amplification, pool balance, library preparation, read assignment, variant calling, intersection logic, individual retesting, and final confirmation. A pilot that stops after a library yield measurement does not test retrieval.

Build the Amplicon Screen

Amplicon selection and sequencing should be sized to the biological targets, not to the maximum multiplex count a platform can advertise. Test each target individually, then in the proposed multiplex context. Inspect product specificity, amplification balance, founder sequence, coverage distribution, strand support, and behavior in representative pools. Targets that consistently underperform should be redesigned, separated, or labeled as limited before production.

The TILLING by NGS service is relevant when many population members or multiple target regions need pooled sequence screening. A smaller targeted sequencing service may fit focused confirmation or a compact target panel. The method choice must preserve the line-retrieval logic; more reads do not repair ambiguous pool membership, non-specific amplification, or a missing seed record.

Define variant-calling rules before inspecting results. At minimum, specify reference and annotation versions, read-quality and mapping filters, minimum usable depth, expected pooled allele-fraction ranges, strand or read-position checks, recurrence filters, and rules for overlapping amplicons. Set thresholds empirically with pilot controls because pool depth, ploidy, zygosity, error profile, and copy specificity change the expected signal. Retain raw counts and quality fields so borderline candidates can be reviewed without rerunning the entire pipeline.

Filter Candidate Mutations

Candidate ranking needs two independent filters: technical credibility and biological relevance. First ask whether the signal is supported by the pool design and sequence evidence. Then annotate the variant against the selected transcript and decide whether its consequence matches the original question. An attractive stop-gain prediction does not rescue a candidate with strand bias, an unbalanced pool intersection, or evidence of paralog misalignment.

For EMS-derived material, transition patterns expected from the mutagen can support review, but they should not become an absolute exclusion rule. Rank nonsense, canonical splice, missense, synonymous, and regulatory candidates according to the study goal. Missense prioritization can consider conserved residues, protein domains, physicochemical change, and multiple prediction tools, while acknowledging that computational effect scores are hypotheses. Independent alleles in the same gene often strengthen a functional study more than a single highly scored candidate.

Preserve rejected calls with explicit reason codes such as insufficient depth, pool inconsistency, recurrent founder allele, ambiguous locus, low-quality position, or confirmation failure. This makes target gaps and pipeline decisions auditable.

Confirm the Individual Line

Confirmation should break the dependence between discovery and verification. Deconvolute the pool to an individual, then test an independent DNA aliquot or, preferably, a new extraction from traceable tissue. Use locus-specific primers that cover the candidate and nearby sequence. Bidirectional Sanger sequencing services can provide a practical orthogonal check for a small number of individual candidates; a larger confirmation set may justify another targeted workflow.

Compare the chromatogram or sequence evidence with the founder and relevant controls. Confirm the genomic coordinate, reference and alternate allele, target copy, zygosity interpretation, and sample identity. If the signal appears only in the original pooled library, classify it as unconfirmed. If an M2 individual is heterozygous, the stored M3 family may segregate. The project should state which plant supplied discovery DNA and which seed lot will be advanced so genotype expectations are not transferred incorrectly across generations.

Confirmation proves the presence of the sequence variant in the tested individual. It does not prove expression change, protein effect, trait causality, environmental stability, or breeding value. Those questions require segregation, independent alleles, complementation or other functional tests, and replicated phenotyping.

The same evidence discipline applies beyond induced-mutant screens. The crop CRISPR validation resource illustrates why sequence detection, locus assignment, sample identity, and biological interpretation should remain distinct validation steps even when the upstream method differs.

Connect Genotype to Seed

A confirmed mutation without viable, correctly labeled seed is a sequence observation rather than a usable allele resource. Reconcile the confirmed DNA ID against packet ID, harvest plant, generation, family, storage location, seed count, and germination history. Reserve enough seed for regeneration and verification; do not consume the only stock in the first phenotype trial. When a line is advanced, assign descendant IDs without erasing the relationship to the discovery plant.

Chemical mutagenesis creates background variants throughout the genome. Recover homozygous descendants where appropriate, genotype the candidate across segregating progeny, and consider backcrossing or independent alleles before attributing a phenotype. Record unexpected sterility, weak seed set, or viability loss as part of allele utility. The 2023 plant TILLING review emphasizes that background mutations remain a central interpretive issue even when target discovery is technically correct.

Define the Delivery Package

The final package should let another researcher reconstruct what was tested and what remains uncertain. Separate measured results from annotations and interpretations. Use versioned, machine-readable tables with stable sample and target identifiers.

Deliverables should include:

• screened and missing line manifest, generation, population, and source-plate position;
• target definition with reference build, transcript, coordinates, primers, and callable intervals;
• pool design and sample-to-pool membership table;
• per-target and per-pool QC, coverage summaries, and failed-target reasons;
• variant table with read evidence, pool intersections, filters, and consequence annotations;
• individual confirmation status, method, trace files or reads, and reviewer decision;
• seed linkage, available generation, zygosity status, and retrieval exceptions;
• methods, software and database versions, and all project-specific thresholds;
• limitations covering untested bases, unresolved paralogs, absent samples, and non-confirmed candidates.

Do not collapse "not detected," "not callable," "not screened," and "not confirmed" into one blank field. These states lead to different next actions and different claims about the population.

Prepare a Quotation Brief

A useful quotation request contains the crop and founder genotype; mutagen and treatment; sampled generation; total and available line counts; DNA format, concentration method, volume, and plate layout; seed inventory; reference assembly and transcript version; target gene list; desired allele classes; paralog concerns; proposed or unknown mutation density; preferred pool strategy; expected confirmation method; and required output files. Include a representative target sequence and a de-identified sample manifest when possible.

Ask the project team to label assumptions. If population size is fixed but mutation density is unknown, price a pilot and staged expansion. If the target list may grow, define the decision date and multiplex redesign policy. If seed retrieval is handled by another institute, state the handoff owner and identifier format. These details have greater impact on schedule and usable results than a target-gene count alone.

Support for TILLING Projects

CD Genomics can support agricultural research teams with target review, pool-aware assay planning, amplicon screening, sequence analysis, and individual candidate confirmation. The TILLING service provides the broader mutation-screening route, while the linked sequencing services can be selected around population scale and confirmation needs. Project specifications are customized to the crop, starting genotype, population structure, target architecture, DNA inventory, and requested evidence state. These services are for agricultural and biological research and are not offered as clinical or diagnostic testing.

TILLING Project FAQ

References

1. Lakhssassi N, Zhou Z, Cullen MA, et al. TILLING-by-Sequencing+ to Decipher Oil Biosynthesis Pathway in Soybeans: A New and Effective Platform for High-Throughput Gene Functional Analysis. International Journal of Molecular Sciences. 2021;22(8):4219.
2. Stupar RM, Locke AM, Allen DK, et al. Soybean genomics research community strategic plan: A vision for 2024–2028. The Plant Genome. 2024;17(4):e20516.
3. Amri-Tiliouine W, Jankowicz-Cieslak J, Till BJ, Laouar M. Generation of Mutant Plants by Gamma Ray Exposure and Development of Low-Cost TILLING Population in Chickpea (Cicer arietinum L.). Methods in Molecular Biology. 2022;2484:143–159.
4. Szurman-Zubrzycka M, Kurowska M, Till BJ, Szarejko I. Is it the end of TILLING era in plant science?. Frontiers in Plant Science. 2023;14:1160695.
5. Kumar J, Kumar A, Sen Gupta D, Kumar S, DePauw RM. Reverse genetic approaches for breeding nutrient-rich and climate-resilient cereal and food legume crops. Heredity. 2022;128(6):473–496.
6. Jiang C, Lei M, Guo Y, et al. A reference-guided TILLING by amplicon-sequencing platform supports forward and reverse genetics in barley. Plant Communications. 2022;3(4):100317.
7. Tsai H, Ngo K, Lieberman M, Missirian V, Comai L. Tilling by sequencing. Methods in Molecular Biology. 2015;1284:359–380.
8. Cooper JL, Till BJ, Laport RG, et al. TILLING to detect induced mutations in soybean. BMC Plant Biology. 2008;8:9.

This content and the described services are intended for agricultural and biological research. They do not provide clinical diagnosis, treatment decisions, or individual health assessment.